CN111968072B - Automatic decision-making method for welding position of thick plate T-shaped joint based on Bayesian network - Google Patents
Automatic decision-making method for welding position of thick plate T-shaped joint based on Bayesian network Download PDFInfo
- Publication number
- CN111968072B CN111968072B CN202010644425.1A CN202010644425A CN111968072B CN 111968072 B CN111968072 B CN 111968072B CN 202010644425 A CN202010644425 A CN 202010644425A CN 111968072 B CN111968072 B CN 111968072B
- Authority
- CN
- China
- Prior art keywords
- welding
- node
- point
- feature points
- filling
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000003466 welding Methods 0.000 title claims abstract description 186
- 238000000034 method Methods 0.000 title claims abstract description 28
- 230000000007 visual effect Effects 0.000 claims abstract description 23
- 238000005070 sampling Methods 0.000 claims abstract description 13
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 10
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000001514 detection method Methods 0.000 claims description 6
- 239000000945 filler Substances 0.000 claims description 6
- 230000035772 mutation Effects 0.000 claims description 5
- 230000037452 priming Effects 0.000 claims description 4
- 238000009499 grossing Methods 0.000 claims description 3
- 230000002401 inhibitory effect Effects 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000004021 metal welding Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000000977 initiatory effect Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 229910000679 solder Inorganic materials 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K37/00—Auxiliary devices or processes, not specially adapted to a procedure covered by only one of the preceding main groups
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K9/00—Arc welding or cutting
- B23K9/32—Accessories
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30152—Solder
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Mechanical Engineering (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Mathematical Optimization (AREA)
- Algebra (AREA)
- Pure & Applied Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Computational Mathematics (AREA)
- Optics & Photonics (AREA)
- Mathematical Analysis (AREA)
- Plasma & Fusion (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The invention provides an autonomous decision method for the initial/welding position of a thick plate T-shaped joint based on a Bayesian network model by means of welding experience and visual conversion characteristics required by welding legs. Firstly, according to the characteristic information of the outline of a welding seam to be welded detected by visual sensing, converting the requirements of welding feet into visual description characteristics, and judging the filling state of the welding seam as three stages of bottoming, filling and cover surface welding; secondly, establishing a Bayesian network model by utilizing the real-time detected weld contour feature points and the weld filling judging state; and finally, combining with an estimated posterior important sampling reasoning algorithm, carrying out real-time decision on the welding position based on the maximum posterior probability criterion, and selecting the welding position with the maximum posterior probability from the identified welding seam contour feature points. The method aims to realize autonomous decision of initial welding positions and welding positions during welding in a plurality of T-shaped joints with web plate thicknesses exceeding 30mm through the algorithm model, and improve welding efficiency.
Description
Technical Field
The invention relates to an autonomous decision-making method for a welding position of a thick plate T-shaped joint based on a Bayesian network, and belongs to the technical field of automatic welding based on visual sensors.
Background
Real-time autonomous decision making of multi-pass arc welding locations in multiple layers is one of the main techniques that affect the welding efficiency of thick plates. At present, multi-layer multi-pass arc welding is still adopted for thick plate welding. In multi-layer multi-pass automatic welding, a certain position needs to be selected from the welding line outline of a region to be welded continuously for multiple times to serve as a welding position in the starting welding and tracking of the next welding line, and the selection process is a welding position decision process.
Because different welding positions can generate different welding effects, the effective welding position decision method not only can improve the welding efficiency, but also can be related to the welding quality. Because effective decision-making of welding positions in multi-channel welding needs to be based on relevant welding knowledge, current groove detection information and the like, research on technologies, algorithm flows and the like related to the effective decision-making have great challenges in implementing autonomous decision-making processes of welding positions. At present, related researches are stopped at welding position detection, welding seam planning based on simulation environment, welding seam preliminary planning based on visual sensing, on-line correction and the like, and no research is involved for deciding a proper welding position in real time. Therefore, by comprehensively considering the working condition specificity of thick plate welding, it is necessary to design a method capable of improving the welding efficiency while meeting the welding quality of the joint in engineering.
Disclosure of Invention
Aiming at the defects of the real-time decision-making method of the welding position of the thick plate T-shaped joint, the invention provides an autonomous decision-making method of the welding position of the thick plate T-shaped joint based on a Bayesian network, and the model is utilized to realize real-time autonomous decision-making of the welding position during multi-path welding initiation/welding, so as to aim at realizing the meeting of the welding quality of the joint and improve the welding efficiency. In order to achieve the above purpose, the technical scheme adopted by the invention is as follows:
a Bayesian network-based autonomous decision method for the welding position of a thick plate T-shaped joint comprises the following steps:
step one, according to the characteristic information of the outline of the welding seam to be welded detected by visual sensing, converting the welding leg requirement into visual description characteristics, and judging the filling state of the welding seam as three stages of bottoming, filling and cover surface welding.
And secondly, establishing a Bayesian network model by utilizing the weld contour feature points and the weld filling judgment states detected in real time, and calculating the conditional probability of each sub-node by combining the empirical knowledge of the welding position and the feature point coordinate information identified in real time.
And thirdly, obtaining posterior probability of each weld contour feature point node by using an estimated posterior important sampling reasoning algorithm, and selecting a corresponding feature point as a welding position of the current sampling based on a maximum posterior probability rule.
Further, in the first step, feature information of the outline of the weld to be welded is detected according to visual sensing, and the specific steps are as follows:
the first step: according to the welding seam image acquired based on the visual sensing of the structured light, the identified welding seam outline is thinned to have only one welding seam data point in the vertical direction, and the slope is calculated by adopting the formula (1):
where x (), y () represent coordinates of a data point; n represents the number of data points involved in each calculation, j represents the subscript of the data point, and i represents the subscript of the data point adjacent to j.
And a second step of: the slope disturbance is further suppressed by using a one-dimensional smoothing filter.
And a third step of: non-linear piecewise fitting of slope data using polynomials up to 50 th order:
f(x)＝a 1 x 50 +a 2 x 49 +…a 50 (2)
wherein the coefficient a 1 ,a 2 ,…a 50 Determined by the coordinates of the fitted data points.
Fourth step: firstly, the fitting result replaces the original slope data to carry out mutation detection. And secondly, acquiring slope monotone sections and positions thereof by adopting a formula (3), marking the lengths of the sections according to the slope change rate of each monotone section, and sequencing the sections from large to small.
Fifth step: finally, the number N of the observation result supervision feature points is introduced, the positions of the feature points are determined by selecting the first N monotone sections in sequence, and the positions of the feature points are determined by the determined monotone sections.
Further, in the first step, the leg requirements are converted into visual description features, and the specific steps are as follows:
the first step: determining an included angle beta between the surface of the welding leg and the web plate according to the required sizes K1 and K2 of the joint welding leg;
and a second step of: identifying characteristic points of a weld contour, determining an upper boundary point A of a web break from the identified characteristic points, then respectively linearly fitting data of the web and data of a bottom plate to obtain two linear equations, and determining an intersection point B of the two linear equations;
and a third step of: and (3) making a straight line passing the point A in the image, so that an included angle between the straight line and a straight line obtained by the fitted web data is beta, and an intersection point of the straight line and the straight line obtained by the fitted bottom plate data is D.
Further, in the first step, the filling state of the welding seam is judged to be three stages of priming, filling and cover welding, and the specific steps are as follows:
the first step: determination ofAnd->Whether or not it is true, in the formula->Ordinate indicating the rightmost feature point,/->Representing the ordinate of the ith feature point, when the rightmost feature point is the lowest in the image, and the rightmost feature point is in the vertical direction with other feature pointsWhen the distance difference is more than or equal to 20pixel, the current welding state is priming welding.
And a second step of: when the right-most feature point identified by the weld profile is to the left of point D and the weld is in non-priming welding, the current weld is in the filler weld stage.
And a third step of: the current welding state is judged to be neither a backing welding stage nor a filling welding stage, and then the current welding state is judged to be a cover welding stage.
Further, the conditional probability calculation of the Bayesian network and each child node for autonomous decision of the welding position comprises the following specific steps:
the first step: the father node of the Bayesian network is in a 'welding state', the child node of the first layer is the horizontal distance between all the identified other feature points and the rightmost feature point, the child node of the second layer is the vertical distance between all the identified other feature points and the rightmost feature point, the child node of the third layer is the identified other feature points, and the last layer is the rightmost feature point.
And a second step of: designating X to represent a parent node variable 'welding state', wherein the parent node variable 'welding state' has two states of 'filling welding' and 'cover welding'; designate Y i (i=1, 2, …) represents a node "horizontal distance i", with two states "satisfied" and "unsatisfied"; designate W j (j=1, 2, …) represents the node "vertical distance j", and there are two states "satisfied" and "unsatisfied"; designate Z k (k=1, 2, …) represents a "feature point k" node, with two states "selected" and "unselected; the designation O represents a "rightmost feature point" node, and there are also two states, "select" and "not select".
And a third step of: the conditional probability of a child node is defined as follows: by usingRepresenting node Y k Conditional probability at Representing node W k Conditional probability>Representing node Z k Conditional probability at, and with pi (Z k ) Representing node Z k Parent node at node O, pi (O) represents parent node at node O, d ij Representing the conditional probability at O.
Fourth step: the "horizontal distance l" and the "vertical distance l" in the model are calculated, and the two variables refer to the horizontal distances (H l ＝x rm -x l ) And vertical distance (V) l ＝y rm -y l )，x rm And y rm The coordinates of the rightmost feature point in the image are respectively.
Fifth step: distance threshold H is respectively set in the horizontal and vertical directions 0 And V is equal to 0 When the distances between other characteristic points and the rightmost characteristic point in the horizontal and vertical directions are all larger than H 0 And V is equal to 0 The feature point is then involved in the weld position decision.
Sixth step: in the filler welding stage:
wherein, x is more than or equal to 3 and k is E (0, 1), and the value is not unique. Experiments verify that x=5, k=0.7 meets the decision requirement.
Seventh step: in the cover welding stage:
wherein,and->
Eighth step: definition of the definitionAnd->Then-> Also, define->And->Thus-> Wherein the operation symbol V represents a logical operation or operation.
Further, the specific steps for implementing the real-time decision of the welding position based on the maximum posterior probability criterion are as follows:
the first step: and determining real-time evidence of the father node, namely determining the stage in which the current welding is positioned.
And a second step of: and obtaining posterior probability of each weld contour feature point node by using an estimated posterior importance sampling reasoning algorithm.
And a third step of: and selecting the characteristic point with the maximum posterior probability as the welding position of the current sampling.
The invention has the beneficial effects that:
the invention provides an autonomous decision method for the welding position of a thick plate T-shaped joint based on a Bayesian network model, which is an innovative and reliable algorithm. And then, a Bayesian network model for multi-pass welding position decision of the T-shaped joint is provided, the effectiveness of the model for implementing the autonomous decision of the welding position is verified, and a foundation is laid for implementing the autonomous decision of the welding position of other typical joints.
Drawings
FIG. 1 is a flow chart of an autonomous decision making method for the welding position of a thick plate T-shaped joint based on a Bayesian network model;
FIG. 2 is a diagram illustrating an exemplary visual characteristic transformation of the weld joint imaging analysis and leg requirements of the T-joint of the present invention;
FIG. 3 is a diagram showing an exemplary structure of a Bayesian network model in accordance with the present invention (four feature points are taken as an example);
FIG. 4 is a graph showing an exemplary decision of the position of filler metal welding in multi-pass welding of the front face of a T-joint with a web thickness of 30mm according to the present invention;
FIG. 5 is a diagram showing an exemplary decision of the welding position of the face welding in the multi-pass welding of the back surface of a T-shaped joint with a web thickness of 30mm according to the present invention;
FIG. 6 is a graph showing an exemplary decision of the position of filler metal welding in multi-pass welding of the back of a T-joint having a web thickness of 50mm in accordance with the present invention;
FIG. 7 is a graph showing an exemplary determination of the position of a face weld in a multi-pass welding of the front face of a T-joint having a web thickness of 50mm in accordance with the present invention.
Detailed Description
The invention will be further illustrated with reference to specific examples, but the invention is not limited to the examples.
The experimental background related to the invention is: the welding method is MAG welding, and the shielding gas is 20% CO 2 +80% Ar, Q345b, 30mm and 50mm web thickness, K-groove for filling, and about 7 and 11 welds for single-side filling. H 0 =20 pixels, V 0 =120 pixels. Experiments verify that x=5, k=0.7 meets the decision requirement.
Example 1: the web thickness was 30mm. A Bayesian network-based autonomous decision method for the welding position of a thick plate T-shaped joint is shown in fig. 1, and comprises the following steps:
step one, detecting characteristic information of a to-be-welded weld line contour according to visual sensing, converting a welding leg requirement into visual description characteristics, and defining a weld line filling state as three stages of bottoming, filling and cover welding.
And secondly, establishing a Bayesian network model by utilizing the weld contour feature points and the weld filling judgment states detected in real time, and calculating the conditional probability of each sub-node by combining the empirical knowledge of the welding position and the feature point coordinate information identified in real time.
And thirdly, obtaining posterior probability of each weld contour feature point node by using an estimated posterior important sampling reasoning algorithm, and selecting a corresponding feature point as a welding position of the current sampling based on a maximum posterior probability rule.
In the first step, feature information of the outline of the weld to be welded is detected according to visual sensing, and the specific steps are as follows:
the first step: from the weld image acquired based on visual sensing of structured light (as in fig. 2 a), the identified weld profile is refined to have and only one weld data point in the vertical direction, and the slope is calculated using equation (1):
where x (), y () represent coordinates of a data point; n represents the number of data points involved in each calculation, the larger the value of n in the range of 2-10 is, the better the effect of inhibiting slope fluctuation is, but the larger the calculation consumption is, n=5 is taken in the study; j represents the subscript of the data point, i represents the subscript of the data point adjacent to j;
and a second step of: further inhibiting slope disturbance by adopting a one-dimensional smoothing filter, and selecting a filter window of 1 multiplied by 9;
and a third step of: and a polynomial with the degree of 50 is adopted to carry out nonlinear piecewise (equipartition) fitting (formula 2) on the slope data, so that the overall change trend of the slope is effectively approximated, and the influence of the local disturbance of the slope on the subsequent mutation detection is reduced.
f(x)＝a 1 x 50 +a 2 x 49 +…a 50 (2)
Wherein the coefficient a 1 ,a 2 ,…a 50 Determining through coordinates of the fitted data points;
fourth step: to detect this mutation, a mutation test is first performed with the fitting result instead of the original slope data. Secondly, acquiring slope monotone sections and positions thereof by adopting a formula (3), marking the length of each section according to the slope change rate of each monotone section, and sequencing the sections from large to small;
fifth step: finally, the number N of the observation result supervision feature points is introduced, the positions of the feature points are determined by selecting the first N monotone sections in sequence, and the positions of the feature points are determined by the determined monotone sections.
The decision of the welding position in the first step is related to the current filling state of the groove, and the filling state of the thick plate joint is generally divided into backing welding, filling welding and cover welding. In order to realize real-time autonomous decision, the state needs to be automatically judged by utilizing the visual characteristics of online detection, and the specific steps are as follows:
the first step: acquiring a backing weld contour by utilizing visual sensing based on structured light (as shown in fig. 2 b), determining a weld contour belonging to a web area according to the identified weld contour feature points, taking the relevant feature point A as a mark to obtain all data points belonging to the web contour, and performing linear fitting on the data points to obtain a linear equation L1;
and a second step of: determining the weld contour belonging to the bottom plate area, taking the relevant characteristic points as marks, obtaining all data points belonging to the bottom plate contour, and performing linear fitting on the data points to obtain a linear equation L2;
and a third step of: determining the intersection point B of the straight lines L1 and L2 (as shown in FIG. 2 d), and determining the intersection angle alpha of the straight lines L1 and L2, wherein the distortion angle between the two plates is as follows
Fourth step: determining the angle beta between the surface of the welding leg and the web plate according to the requirement K1 from the upper end A of the web plate breach to the bottom plate and the requirement K2 of the bottom plate (as shown in figure 2 c);
fifth step: making a straight line L3 in the image by passing A, and enabling an included angle between L3 and L1 to be beta;
sixth step: determining an intersection point D of L3 and L2 (as shown in FIG. 2D), wherein a line segment BD in the image represents the requirement of the solder fillets on the bottom plate;
seventh step: the perpendicular to line segment AD is drawn through point B, the foot is C, and the line segment BC is rotated by gamma degrees clockwise to BC ', BC' representing the 3 rd requirement of the fillets: the web extends a distance from the foot surface at the intersection of the bottom plates.
Finally, judging the current welding state according to the positions of the rightmost characteristic point and the point D which are recognized in real time, and firstly judgingAnd->Whether or not to meet, wherein->Ordinate indicating the rightmost feature point,/->Representing the ordinate of the ith feature point, and if both feature points are satisfied, performing backing welding on the current welding state; if the welding is not backing welding and the rightmost feature point is on the left side of the point D, the current welding is filling welding; if the welding is neither backing welding nor filling welding, the current welding is cover welding; and if the judgment result is the backing welding, the welding position is the average coordinate position of the second and third characteristic points.
The conditional probability calculation of the Bayesian network model and the child nodes in the second step is as follows:
a bayesian network model of autonomous decision of weld location for the filler and cap welding phases is shown in fig. 3.
Firstly, judging the welding state of the current welding to obtain evidence of a father node;
next, except for the leftmost feature point, the horizontal distance H between other feature points and the rightmost feature point is calculated in turn l ＝x rm -x l And vertical distance V l ＝y rm -y l Wherein x is rm And y rm Coordinates of the rightmost feature points in the image;
thirdly, calculating the conditional probability of each sub-node of the first layer, and in the filling welding stage:
wherein, x is more than or equal to 3 and k is E (0, 1), and the value is not unique. Experiments verify that x=5, k=0.7 meets the decision requirement.
In the cover welding stage:
wherein,and->Definitions->And is also provided withThen->Also, define->And->Thus->Wherein the operation symbol V represents a logical operation or operation.
The specific steps of implementing the welding position real-time decision based on the maximum posterior probability criterion in the third step are as follows:
firstly, after determining prior probability and real-time evidence nodes, the model obtains posterior probability of each target node (weld contour feature point) by using an estimated posterior important sampling reasoning algorithm;
and secondly, selecting the characteristic point with the maximum posterior probability as the welding position of the current sampling.
Firstly, one-time welding of front welding of a joint in a filling welding stage is selected as a test object, the profile morphology of a welding seam collected before and during welding is shown in fig. 4a, according to the obtained characteristic point information (fig. 4 b) of the profile of the welding seam, decision results of a Bayesian network and a hierarchical analysis method proposed in the text are both the characteristic point 1 as a welding position, corresponding decision basis is different from a process (the left part of fig. 4 c), and the welding result is shown in fig. 4d. And secondly, selecting one-time welding of welding the back surface of the joint in the cover surface welding stage, wherein visual characteristic information of a welding line is shown in fig. 5a and b, the welding position decided by the two methods is taken as a characteristic point 2, the decision is shown in the right part of fig. 4c, and the welding result is shown in fig. 5c.
Example 2: the web thickness was 50mm and the weld position decision process was performed similarly to example 1.
Test results show that for the problem of selecting multi-layer multi-channel welding positions with the thickness of the T-shaped joint web plate within 50mm, the Bayesian network decision model provided by the invention can meet autonomous decision, and provides technical support for further realizing intelligent welding of thick plates; the welding position decision can be effectively implemented by using the analytic hierarchy process based on the maximum probability rule, but the problem is that a judgment matrix cannot be effectively and automatically constructed so as to automatically adapt to the change of the number of the characteristic points of the welding line profile.
The foregoing description of the preferred embodiments of the present invention has been presented only in terms of those specific and detailed descriptions, and is not, therefore, to be construed as limiting the scope of the invention. It is specifically stated that the above-described calculation method for the prior probability is not unique but is still included in the scope of the present invention.
Claims (5)
1. A Bayesian network-based autonomous decision method for the welding position of a thick plate T-shaped joint is characterized in that: the method comprises the following steps:
step one, detecting characteristic information of a to-be-welded weld contour according to visual sensing, converting a welding leg requirement into visual description characteristics, and judging a weld filling state as three stages of bottoming, filling and cover welding;
establishing a Bayesian network model by utilizing the weld contour feature points and the weld filling judgment states detected in real time, and calculating the conditional probability of each sub-node by combining the empirical knowledge of the welding position and the feature point coordinate information identified in real time;
the conditional probability calculation of the Bayesian network and each child node in the second step comprises the following specific steps:
the first step: the father node of the Bayesian network is in a 'welding state', the first layer of child nodes are horizontal distances between all the identified other feature points and the rightmost feature points, the second layer of child nodes are vertical distances between all the identified other feature points and the rightmost feature points, the third layer of child nodes are the identified other feature points, and the last layer of child nodes is the rightmost feature points;
and a second step of: designating X to represent a parent node variable 'welding state', wherein the parent node variable 'welding state' has two states of 'filling welding' and 'cover welding'; designate Y i (i=1, 2, …) represents a node "horizontal distance i", with two states "satisfied" and "unsatisfied"; designate W j (j=1, 2, …) represents the node "vertical distance j", and there are two states "satisfied" and "unsatisfied"; designate Z k (k=1, 2, …) represents a "feature point k" node, with two states "selected" and "unselected; designating O to represent a node of a right-most feature point, and also having two states of "select" and "not select";
and a third step of: the conditional probability of a child node is defined as follows: by usingRepresenting node Y k Conditional probability> Representing node W k Conditional probability> Representing node Z k Conditional probability at, and with pi (Z k ) Representing node Z k Parent node at node O, pi (O) represents parent node at node O, d ij Representing a conditional probability at O;
fourth step: the "horizontal distance l" and the "vertical distance l" in the model are calculated, and the two variables refer to the horizontal distances (H l ＝x rm -x l ) And vertical distance (V) l ＝y rm -y l )，x rm And y rm Coordinates of the rightmost feature points in the image;
fifth step: distance threshold H is respectively set in the horizontal and vertical directions 0 And V is equal to 0 When the distances between other characteristic points and the rightmost characteristic point in the horizontal and vertical directions are all larger than H 0 And V is equal to 0 When the characteristic point is in the welding position decision, the characteristic point participates in the welding position decision;
sixth step: in the filler welding stage:
wherein, x is more than or equal to 3 and k is E (0, 1), and the value is not unique;
seventh step: in the cover welding stage:
wherein,and->
Eighth step: definition of the definitionAnd->Then-> Also, define->And->ThenWherein the operation symbol V represents a logical operation or operation;
and thirdly, obtaining posterior probability of each weld contour feature point node by using an estimated posterior important sampling reasoning algorithm, and selecting a corresponding feature point as a welding position of the current sampling based on a maximum posterior probability criterion.
2. The autonomous decision making method for the welding position of a thick plate T-joint according to claim 1, wherein: in the first step, the characteristic information of the outline of the weld to be welded is detected according to visual sensing, and the specific steps are as follows:
the first step: according to the welding seam image acquired based on the visual sensing of the structured light, the identified welding seam outline is thinned to have only one welding seam data point in the vertical direction, and the slope is calculated by adopting the formula (1):
where x (), y () represent coordinates of a data point; n represents the number of data points involved in each calculation, j represents the footer of the data point, i represents the footer of the data point adjacent to j;
and a second step of: further inhibiting slope disturbance by adopting a one-dimensional smoothing filter;
and a third step of: non-linear piecewise fitting of slope data using polynomials up to 50 th order:
f(x)＝a 1 x 50 +a 2 x 49 +…a 50 (2)
wherein the coefficient a 1 ,a 2 ,…a 50 Determining through coordinates of the fitted data points;
fourth step: firstly, carrying out mutation detection on fitting results instead of original slope data, secondly, obtaining slope monotone sections and positions thereof by adopting a formula (3), marking the length of each section according to the slope change rate of each monotone section, and sequencing the sections from large to small;
fifth step: finally, the number N of the observation result supervision feature points is introduced, the positions of the feature points are determined by selecting the first N monotone sections in sequence, and the positions of the feature points are determined by the determined monotone sections.
3. The autonomous decision making method for the welding position of a thick plate T-joint according to claim 1, wherein: in the first step, the welding leg requirements are converted into visual description characteristics, and the specific steps are as follows:
the first step: determining an included angle beta between the surface of the welding leg and the web plate according to the required sizes K1 and K2 of the joint welding leg;
and a second step of: identifying characteristic points of a weld contour, determining an upper boundary point A of a web break from the identified characteristic points, then respectively linearly fitting data of the web and data of a bottom plate to obtain two linear equations, and determining an intersection point B of the two linear equations;
and a third step of: and (3) making a straight line passing the point A in the image, so that an included angle between the straight line and a straight line obtained by the fitted web data is beta, and an intersection point of the straight line and the straight line obtained by the fitted bottom plate data is D.
4. The autonomous decision making method for the welding position of the T-shaped joint of a thick plate according to claim 3, wherein: in the first step, the filling state of the welding line is judged to be three stages of priming, filling and cover surface welding, and the specific steps are as follows:
the first step: determination ofAnd->Whether or not it is true, in the formula->Representing the rightmost feature pointOrdinate of>Representing the ordinate of the ith feature point, and when the rightmost feature point is at the bottommost position in the image and the distance between the rightmost feature point and other feature points in the vertical direction is more than or equal to 20 pixels, performing prime welding in the current welding state;
and a second step of: when the characteristic point at the rightmost end identified by the weld contour is at the left side of the point D and the welding is in non-backing welding, the current welding is in a filling welding stage;
and a third step of: the current welding state is judged to be neither a backing welding stage nor a filling welding stage, and then the current welding state is judged to be a cover welding stage.
5. The autonomous decision making method for the welding position of a thick plate T-joint according to claim 1, wherein: the specific steps for implementing the real-time decision of the welding position based on the maximum posterior probability criterion in the third step are as follows:
the first step: determining real-time evidence of the father node, namely determining the stage of the current welding;
and a second step of: obtaining posterior probability of each weld contour feature point node by using an estimated posterior importance sampling reasoning algorithm;
and a third step of: and selecting the characteristic point with the maximum posterior probability as the welding position of the current sampling.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010644425.1A CN111968072B (en) | 2020-07-07 | 2020-07-07 | Automatic decision-making method for welding position of thick plate T-shaped joint based on Bayesian network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010644425.1A CN111968072B (en) | 2020-07-07 | 2020-07-07 | Automatic decision-making method for welding position of thick plate T-shaped joint based on Bayesian network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111968072A CN111968072A (en) | 2020-11-20 |
| CN111968072B true CN111968072B (en) | 2024-04-02 |
Family
ID=73360894
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010644425.1A Active CN111968072B (en) | 2020-07-07 | 2020-07-07 | Automatic decision-making method for welding position of thick plate T-shaped joint based on Bayesian network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111968072B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112975190B (en) * | 2021-03-23 | 2022-07-08 | 北京石油化工学院 | Multilayer multi-pass welding method, device, equipment and system based on visual sensing |
| CN113762400B (en) * | 2021-09-13 | 2023-10-31 | 南昌大学 | Autonomous extraction method for weld joint position based on naive Bayes classifier |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101526480A (en) * | 2009-04-27 | 2009-09-09 | 清华大学 | Real-time detection method of butt weld of thin plates based on visual sense |
| CN102292187A (en) * | 2008-11-21 | 2011-12-21 | 普雷茨特两合公司 | Method and device for monitoring a laser machining operation to be performed on a workpiece and laser machining head having such a device |
| CN106952281A (en) * | 2017-05-15 | 2017-07-14 | 上海交通大学 | A kind of method that weld profile feature recognition and its welding bead are planned in real time |
| CN109949346A (en) * | 2019-03-04 | 2019-06-28 | 天津科技大学 | Welding seam tracking method based on histograms of oriented gradients particle filter |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| IT1397985B1 (en) * | 2010-02-08 | 2013-02-04 | Prima Ind Spa | MONITORING PROCEDURE FOR THE QUALITY OF LASER PROCESSING PROCESSES AND ITS SYSTEM |
-
2020
- 2020-07-07 CN CN202010644425.1A patent/CN111968072B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102292187A (en) * | 2008-11-21 | 2011-12-21 | 普雷茨特两合公司 | Method and device for monitoring a laser machining operation to be performed on a workpiece and laser machining head having such a device |
| CN101526480A (en) * | 2009-04-27 | 2009-09-09 | 清华大学 | Real-time detection method of butt weld of thin plates based on visual sense |
| CN106952281A (en) * | 2017-05-15 | 2017-07-14 | 上海交通大学 | A kind of method that weld profile feature recognition and its welding bead are planned in real time |
| CN109949346A (en) * | 2019-03-04 | 2019-06-28 | 天津科技大学 | Welding seam tracking method based on histograms of oriented gradients particle filter |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111968072A (en) | 2020-11-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111968072B (en) | Automatic decision-making method for welding position of thick plate T-shaped joint based on Bayesian network | |
| CN109191459B (en) | Automatic identification and rating method for continuous casting billet macrostructure center segregation defect | |
| CN109272544B (en) | Structured light three-dimensional measurement model of pipeline all-position welding line and image processing method | |
| CN106952281B (en) | The method that a kind of identification of weld profile feature and its welding bead are planned in real time | |
| CN109509181B (en) | Cladding pool shape visual detection method based on serial contour search | |
| CN114170176B (en) | Automatic detection method for welding seam of steel grating based on point cloud | |
| Khanzadeh et al. | A methodology for predicting porosity from thermal imaging of melt pools in additive manufacturing thin wall sections | |
| CN112053376B (en) | Workpiece weld joint identification method based on depth information | |
| CN110717872A (en) | Method and system for extracting characteristic points of V-shaped welding seam image under laser-assisted positioning | |
| CN113828947B (en) | BP neural network laser welding seam forming prediction method based on double optimization | |
| CN114240944B (en) | Welding defect detection method based on point cloud information | |
| JP6905956B2 (en) | Crack change determination device and method | |
| CN108921828B (en) | Method for identifying insignificant weld joint in complex scene | |
| Yu et al. | Vision-based deviation extraction for three-dimensional control in robotic welding with steel sheet | |
| CN113674206A (en) | Extraction method suitable for characteristic parameters of deep-melting K-TIG welding molten pool and keyhole entrance | |
| CN106875430A (en) | Single movement target method for tracing and device based on solid form under dynamic background | |
| JP2008267943A (en) | Crack detecting method | |
| CN111489363A (en) | Method for measuring weld beading size of root two-dimensional image of small-diameter pipe weld | |
| CN114119504A (en) | Automatic steel part welding line detection method based on cascade convolution neural network | |
| CN114066752A (en) | Line-structured light skeleton extraction and burr removal method for weld tracking | |
| CN110608684B (en) | Single-layer multi-channel weld accumulation deposition effect detection method and system | |
| Xia et al. | Real-time seam tracking during narrow gap GMAW process based on the wide dynamic vision sensing method | |
| CN111438460B (en) | Vision-based thick plate T-shaped joint welding seam forming characteristic online measurement method | |
| JP4734176B2 (en) | Wire abnormality detection method, wire abnormality detection device, and wire abnormality detection program | |
| CN113762400A (en) | Weld joint position autonomous extraction method based on naive Bayes classifier |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |